What Is Per-Agent Pacing in an Outbound Campaign?

Per-agent pacing is the rule that gives each human campaign agent an independent wait period before that person becomes eligible for another lead. Agent A’s clock can expire while Agent B is still talking or completing a recovery period. This differs from global batching, where the dialer releases a fixed group of calls on one shared timer regardless of which people are ready. In MYLO, pacing should be enforced by deterministic campaign services together with agent availability, capacity, and connection-flow state. It is not an LLM decision made between live calls. The business chooses an appropriate pace for its workflow and people, while MYLO prevents new work from being assigned before the configured lane is eligible.

An Independent Clock for Each Agent

A multi-agent campaign can be understood as several independent lanes. Each lane has an agent, current call state, last completed outcome, and next eligible time. When one lane finishes, its pacing interval begins without changing the clocks of agents who finished earlier or remain active.

The scheduler checks the lane’s current state before claiming another lead. Time alone is not enough: the agent must also be eligible, reachable, unpaused, and supported by caller-ID and channel capacity. This prevents a timer from overriding actual availability.

Why Global Batching Is Different

A global batch might start five attempts every minute because five agents are assigned. If two calls last longer than a minute and one agent pauses, the next batch can create more customer calls than the ready team can handle. Customer-first flows are especially sensitive to this mismatch.

Per-agent pacing admits at most one new unit of work for each lane when that lane is ready. It follows current state rather than the original team size. The campaign can still operate concurrently, but concurrency emerges from several eligible lanes instead of an uncontrolled shared burst.

Agent Recovery and Wrap Time

Agents may need time after a call to enter notes, update a case, review an order, take a short recovery period, or complete a required disposition. Pacing makes that space part of the operating rule rather than depending on agents to race a dialer.

The right interval depends on the work. A short confirmation campaign and a sensitive support callback should not automatically use the same value. Management must balance customer experience, staff wellbeing, data quality, and throughput. MYLO provides the mechanism within supported constraints; it does not set employment policy.

Connection Flow Affects the Customer Risk

In an agent-first flow, the agent leg is established before the customer call. A newly eligible lane can confirm that the person is ready before contacting the lead. In a customer-first flow, the customer answers first and the agent leg follows, so accurate availability and conservative admission are critical.

Pacing does not replace capacity reservation or flow-specific failure handling. The scheduler still needs the required channels, caller ID, endpoint or mobile, and lead ownership. If the agent leg fails, the system should classify that separately from customer no-answer.

A Three-Agent Timeline

At 10:00, three agents are eligible and each lane claims one lead. Agent A’s call ends at 10:04, Agent B’s at 10:06, and Agent C’s at 10:10. With a two-minute pacing interval, their next eligibility times are 10:06, 10:08, and 10:12 respectively. The system does not wait for one global 10:12 release.

If Agent B pauses at 10:07, that lane remains ineligible at 10:08 despite its clock. If Agent A begins another call at 10:06, the next clock begins only after that flow reaches the defined terminal state. This state model keeps behaviour explainable to operators.

Pacing After Different Outcomes

The campaign specification should define which terminal events start pacing. A completed conversation, customer no-answer, provider rejection, agent-side failure, and cancelled attempt may require different operational handling. The system should not infer a new policy from each error message.

Customer retry eligibility is a separate decision. Waiting before an agent receives another lead does not mean the same customer should be called again. A retry should follow the configured attempt policy, schedule, consent, and final-status rules.

Capacity and Pacing Work Together

When a lane becomes time-eligible, a call can start only if the necessary resources are available. Mobile-agent flows may need two external channels, extension flows may need one external customer channel, and other active services may consume shared capacity. Caller-ID assignments must also remain valid.

Conversely, free channels do not justify a call when no agent lane is eligible. The scheduler respects the intersection of people, time, identity, channels, lead state, campaign status, and appliance limits. This prevents one abundant resource from hiding a shortage elsewhere.

Pause, Resume, and Restart

Pausing a campaign should stop admission of new work while handling active calls according to defined policy. Resume should not erase every agent’s pacing history or release a large batch. Each lane continues from reconciled state and its next eligible time.

After an appliance restart, MYLO should reconcile stored attempts with live evidence before treating agents and leads as free. Durable pacing data prevents an interruption from causing duplicate or unexpectedly rapid calls. Operators should see when recovery is complete and why a lane remains unavailable.

Practical Example: Follow-Up Team

Four agents handle authorised customer follow-ups. The business sets a three-minute recovery period because each person must record notes and confirm the next action. Calls end at different times, so each agent returns to eligibility independently. A supervisor can pause one agent for coaching without stopping the other lanes.

The team tests agent-first connection, customer no-answer, agent unavailable, pause, resume, and restart. It confirms that no customer is called simply because another agent’s clock expired and that notes can be completed before reassignment. The business reviews quality and staff feedback before changing the interval.

MYLO and Customer Responsibilities

MYLO deterministically stores and enforces supported campaign and lane state. Asterisk executes calls, and the provider supplies channels and external delivery. The LLM can explain configuration or gather a pacing requirement but should not control the runtime hot loop.

The customer owns campaign purpose, staffing, training, scripts, schedules, employment practices, consent and suppression, and the decision about an appropriate interval. Pacing cannot make an unlawful list acceptable or compensate for inadequate provider capacity and unavailable agents.

Per-Agent Pacing Checklist

• Define the recovery period for the specific campaign workflow.
• Maintain an independent next-eligible time for each agent lane.
• Require current agent eligibility in addition to time eligibility.
• Respect connection-flow channel and caller-ID requirements.
• Define how each terminal attempt result affects pacing.
• Keep customer retry policy separate from agent recovery.
• Pause new admission without losing active lane state.
• Reconcile pacing and attempts after restart.
• Test staggered completions and agent-specific pauses.
• Review quality, customer experience, and staff wellbeing before adjustment.

Per-agent pacing makes a multi-agent campaign behave like coordinated individual lanes. It protects customer experience and agent readiness without forcing the entire team onto one artificial clock.

Choose and Review the Pacing Interval

Observe the work agents must complete after a representative call: disposition, notes, customer record update, follow-up scheduling, and personal recovery. Set an initial interval that accommodates this work without forcing inaccurate shortcuts. Different campaign purposes or agent groups may need different approved values, but the current product’s supported configuration remains the authority.

Review evidence after a controlled period. Signs that pacing is too short include incomplete notes, repeated pauses, missed agent legs, customer hold, and quality decline. Signs that it may be longer than needed include sustained eligible idle time when leads and every other resource are available. Do not optimise solely for attempts per hour.

Make changes safely

A pacing change should identify the campaign, old value, new value, reason, approver, and effective time. Decide whether it applies to already-running clocks or only to future terminal events; ambiguous mid-clock changes can surprise agents. Test pause, resume, staggered completions, and restart after the change.

Keep metrics honest

Report active conversation time, pacing time, paused time, resource-blocked time, and genuinely idle eligible time separately where supported. Combining them as “agent idle” can blame people for provider or capacity constraints. Pacing is a control for safe workflow, not a hidden productivity judgement.

Functional Test for Pacing

Use at least two agent lanes with staggered call endings. Confirm that each next-eligible time is calculated independently, that an early-finishing agent can proceed without releasing the others, and that a paused or unreachable agent remains blocked after its timer expires. Verify that customer retry rules do not change merely because the agent clock changes.

Pause and resume the campaign, then perform a controlled restart and reconcile state. The accepted result should show no burst, duplicate lead claim, lost recovery interval, or new call without the required channels and caller ID. Record timestamps so the behaviour is objectively reviewable.

The Current Human-Operator Default

Current MYLO human campaigns use a default pacing interval of four seconds per human operator. This is a configurable product default, not a telecom law and not one global delay shared by every operator. Customer-to-IVR campaigns have no human operator and therefore do not inherit human-agent pacing.